最开始考虑的自然是SourceForge。但在公叔R要通过Proxy讉Ksourceforge.netӞ速度奇慢?br />
W二个考虑的是Google Code。无论在安过ADSL讉KQ还是在公司通过Proxy讉KQ速度都可以接受。最好用的是支持对Issue, Discussion, Document{自定义属性标{，如priority, type, milestone{?br />
无论是SourceForgeQ还是Google CodeQ与subversion和eclipse的集成都非常好。可是我用的是Microsoft Visual Studio 2008和C#Q与q两个东安不合拍：(x)
- Subversion 与VS2008的集成需要购买Visual SVN
- VS2008不能发布下蝲包到SourceForge和Google Code上?/li>
最后我发现CodePlexq个|站Q居然提供Visual Studio Team Fundation Server的服务。与我用的VS2008完全是一Ӟ可惜的是它的|速非怸E_Q很难访问。估计服务器只在国。最要命的是Q在公司的Proxy环境下，我可以从IE讉KCodePlex的VSTS服务器，但VS2008的Team Explorer怎么也访问不了?/p>
| HBM Element | Java Interface | Java Implementation |
| <set> | Set | HashSet |
| <set> with order | SortedSet | TreeSet |
| <list> | List | ArrayList |
| <bag>, <idbag> | Collection | ArrayList |
| <map> | Map | HashMap |
| <map> with order | SortedMap | TreeMap |
| <array>, <primitive-array> | N/A | array |
Sample Tables
CREATE TABLE `core_sample_company` (
`companyId` decimal(18,0) NOT NULL,
`companyName` varchar(128) NOT NULL,
`description` varchar(1024) default NULL,
PRIMARY KEY (`companyId`)
);CREATE TABLE `core_sample_role` (
`roleId` decimal(18,0) NOT NULL,
`roleName` varchar(128) NOT NULL,
`companyId` decimal(18,0) NOT NULL,
`description` varchar(1024) default NULL,
PRIMARY KEY (`roleId`)
);
CREATE TABLE `core_sample_user` (
`userId` decimal(18,0) NOT NULL,
`userName` varchar(128) NOT NULL,
`companyId` decimal(18,0) NOT NULL,
`defaultRoleId` decimal(18,0) default NULL,
`description` varchar(1024) default NULL,
PRIMARY KEY (`userId`)
);
CREATE TABLE `core_sample_user_role` (
`userId` decimal(18,0) NOT NULL,
`roleId` decimal(18,0) NOT NULL,
`pripority` int(11) NOT NULL,
PRIMARY KEY (`userId`,`roleId`)
);
HBM defintion
The definiton of <set>, <bag>, <list> is similar.
Defines a collection whose element type is simple data type.
<class name="SampleCompany" table="core_sample_company">
<bag name="roleNames" table="core_sample_role" lazy="false" >
<key column="companyId"/>
<element column="roleName" type="string"/>
</bag>
</class>
Query HQL: select c.id, c.name, r from SampleCompany c left join c.roleNames r
Defines a collection whose element type is another mapped java class
<class name="SampleCompany" table="core_sample_company">
<bag name="roles" cascade="none">
<key column="companyId"/>
<one-to-many class="SampleRole" not-found="ignore"/>
</bag>
</class>
Query HQL: select c.id, c.name, r.name from SampleCompany c left join c.role r
Pay attention that key column is a foreign column of SampleRole table.
Defines a list with list-index
<list> is not a popular element. It request a index column in table. The index column is the index of java List, it has to be a sequence starts from 0.
<class name="SampleUser" table="core_sample_user">
<list name="roles" table="core_sample_user_role" cascade="all" lazy="false" >
<key><column name="userId" sql-type="integer"/></key>
<index column="priority"></index>
<many-to-many class="SampleRole">
<column name="roleId"></column>
</many-to-many>
</list>
</class>
The benifit of <list> is it alwasy sorts list by index column. However, It is hard to resort the list. I tried remove a role from role list and add it to another poisition. When save the role list, an exception throwed:
java.sql.BatchUpdateException: Duplicate entry 'user001-role003 for key 1
This should be a hibernate bug.
Defines a bag with relationship table
<class name="SampleRole" table="core_sample_role">
<bag name="users" table="core_sample_user_role" cascade="none" lazy="false">
<key><column name="roleId" sql-type="integer"/></key>
<many-to-many class="SampleUser">
<column name="userId"></column>
</many-to-many>
</bag>
</class>
- Key column is foreign column from relationship table to current table(SampleRole>
- many-to-many sub column is foreign column from relationship table to target table (SampleUser)
]]>
public class PojoModelTree extends IdentifiableTree<? extends PojoModel>
错误信息如下Q?/p>
The type PojoModelTree cannot extend or implement IdentifiableTree<? extends PojoModel>. A
supertype may not specify any wildcard
2、只?lt;? super Type>Q没?lt;E super Type>
正确用法QTreeNode<? super Node> getTree()
错误用法 <E super Node> E getParent();
3、函数返回值类型不应用通配W：(x)
错误用法QTreeNode<? extends T> getChildNode(String pKey)
q种用法本n无错Q但在赋q回值给其它变量时会(x)报类型不匚w?/p>
正确用法Q?lt;N extends T> TreeNode<N> getChildNode(String pKey)
4、带通配W的泛型集合不能使用addҎ(gu)。不带通配W的泛型集合也可接收子类元素?/p>
错误用法Q?/p>
List<? extends Number> list1 = new ArrayList<Number>();
list1.add(new Integer(11)); // cd不匹配?
W二行报错ؓ(f)Q?
The method add(capture#1-of ? extends Number) in the type List<capture#1-of ? extends
Number> is not applicable for the arguments (Integer)
正确用法Q?
List<Number> list1 = new ArrayList<Number>();
list1.add(new Integer(11));
昄Q在泛型的检查之下，仍可向集合中d指定泛型的子cd素。以下代码也是合法的Q?
List<Number> list1 = new ArrayList<Number>();
list1.add(new Integer(11));
List<Integer> list2 = new ArrayList<Integer>();
list2.add(33);
list2.add(44);
list1.addAll(list2);
但是Q如果写list1=list2׃合法了?
<bean id="mySessionFactory" class="org.springframework.orm.hibernate3.LocalSessionFactoryBean">
<property name="dataSource" ref="myDataSource"/>
<property name="mappingResources">
<list>
<value>product.hbm.xml</value>
</list>
</property>
<property name="hibernateProperties">
<value>
hibernate.dialect=org.hibernate.dialect.HSQLDialect
</value>
</property>
</bean>
然而，在我的案例中Q因Z务需要，我们必须在运行时加入新的HBM。在单独使用Hibernate的时Q只要取到Hiberante Configuration对象Q修改一下配|，然后rebuild session factory可以了?
可是lSpringq一装Q我只能取到一个只ȝSessionFactory对象Q无法进行重|?
q里有两U可能方法，但我都不知道如何做?
1Q取到LocalSessionFactoryBean对象q行重置?
2Q取到Hibernate Configuration对象q行重置?
]]>
]]>
http://www.wujianrong.com/archives/2006/11/jdbc.ht...
]]>
q个功能是动态O/R Mapping。就是说Q在pȝq行Ӟ动态生成一份新的O/R MappingQ或者往原来的O/R Mapping中添加字Dc(din)ƈ且这些改动，应当无须重启Application Serverp生效?/p>
Hibernate虽然也没有很好的支持q一点，但是提供了可实现的渠道?/p>
1. 重徏SessionFactory时可以添加新的Mapping文g?/p>
2. Dynamic Component可以在pojo中将动态字D表达ؓ(f)MapQ这样只用修改Mapping文gQ不用修改javacR?/p>
3. Hibernate 3.2.1q提供不用javacȝMapping方式。在Mapping文g中，只指定entity-nameQ不指定classQHiberante׃(x)使用Map来表辑֮?/p>
Jpox没有提供上面三种渠道中的M一U。我已经在Jpox forum上发贴问了，不知道什么时候能有答复?/p>
Kodo的文档中也没有直接提刎ͼ本想去bea发脓(chung)的，但是|速太慢?/p>
我说的完整的RBACQ是指支持角色树(wi)形结构和角色分组。具体来_(d)应当包含如下权限控制需求：(x)
- 父角色可以讉K甚至是修改其子的数据，包含直接子直到最l子U?
- 角色可以讉K其所在组的数据?
- 父角色可以讉K其所有子U?从直接子U到最l子U?所在组的数据?/li>
而具体到我的pȝ中，q应当有如下需求?/p>
- 兼容多种数据库品。只能用单的表，视图Q存储过E和函数{实现?
- 同时兼容单条数据处理和批量数据处理的需求?/li>
且不些具体需求，RBAC的基本表应当如下四个Q?/p>
- roleList表，记录所有的角色和角色组?
- roleId: PK, 角色/l的IDQ全局唯一Q不区分角色和组?
- roleName:角色/l的名称?
- roleType: R - 角色QG - l?/li>
- rolePermission表，记录每一个角?l对每一个对象的权限?
- permissionID: PK, 无特定意义?
- role: 角色/l的ID?
- object: 对象的ID?
- permission: 权限标识Q如读，写，删等?/li>
- roleRelationship表，记录角色/l之间的关系?
- relationId: PK, 无特定意义?
- superiorRole: 父角?l的ID?
- role:子角Ԍ子组Q成员角Ԍ成员l的ID?
- relationship: 关系标识Q可在如下设|集中选取一个?
- PG标识QP - 父子关系QG - l?成员关系?
- PPGG标识Q在PG集上Q再加三U：(x)PP - 间接父关系QGG - l内l关p，CG - parentRole是组QchildRole的子角色或间接子角色是其成员Q或其子l?含间接子l）的成?/li>
- objectList表，记录所有的对象?
- objectId: PKQ对象IDQ全局唯一?
- objectName: 对象名称?
- ... ...
分析上述表结构，不难发现Q问题的关键在于从rolePermission表中d数据Ӟ如何限定角色/l的范围.
Ҏ(gu)一
如果角色和组的总量不大Q比如在100以内Q采用PPGG标识关系Q读取数据时是最快的。这个时候的SQL只需要一个输入参?roleIdQ?/p>
SELECT object FROM rolePermission p left join roleRelationship r on p.role = r.role WHERE p.role = ?roleId or r.superiorRole = ?roleId. (未验证SQL的正?
但是Q这个方案是以极度冗余roleRelationship表的数据ZL(fng)Q比如有100个角Ԍ那么roleRelationship中将?x)?00 * 100 Q?0,000条记录。而在每次调整角色和R角色l的时候，p在roleRelationship中一ơ增加或删除100条记录。这个开销是比较大的?/p>
Ҏ(gu)?/p>
只标识PGQ查询时接收的输入参Cؓ(f)一个完整的相关角色列表?roleList?/p>
SELECT object FROM rolePermission WHERE role in (?roleList)
在系l运行时Q这?roleList通常可以从role hierarchy cache中取刎ͼ比较方便。这个方案的主要问题有二Q?/p>
1Q如?roleListq长Q用in判断性能?x)很差?/p>
2Q在有些情况下，如报表查询和pȝ外查询时Q取得roleList不太方便?/p>
Ҏ(gu)?/p>
只标识PGQ但使用如下三个数据库函数来判断角色/l之间的关系?/p>
- boolean isChild(role, parentRole) - 如role为parentRole的子Q返回true?
- boolean isDescendant(role, ancestorRole) - 如role为ancestorRole的子或间接子U，q回true?
- boolean isMember(role, group) - 如role为group的成员或子组的成员，q回true?
- boolean descendantIsMember(role, group) - 如role的子或间接子Uؓ(f)group的成员，q回true?
- boolean isBelong(role, super) - 如role为super的子Q间接子Q成员或间接员，或者role的子(含间接子)是super的成员或子组成员Q返回true?/li>
在查询时Q也只需要接收一?roleIdQSELECT object FROM rolePemission WHERE isBelong(?roleId, role)
如何写出高性能的数据库函数是实现这个方法的关键?/p>
上述Ҏ(gu)仅是理论分析Q我們于方案二?/p>
l于惛_新的Ҏ(gu)了?/p>
Ҏ(gu)四，
l合Ҏ(gu)一和方案二Q在roleRelationship中，对前两(也可以是三或四U?角色Q保存其所有的下角色和组。这P如果以前两角色查询数据Q就使用Ҏ(gu)一Q如果以W三U及(qing)以下的角色查询数据，׃用方案二?/p>
仍以100个角色ؓ(f)例，每个角色要保存三个关p：(x)一U主角Ԍ二ȝ角色Q直接主角Ԍ最多有300条数据?/p>
每往角色l中加一个角Ԍ也需要加入三条数据：(x)角色本nQ一U主角Ԍ二ȝ角色?/p>
但往角色l中加一个子l，需要加入的数据量就大一些：(x)子组本nQ子l所有角Ԍ子组所有角色的一U主角色和二ȝ角色。如在多个子l中发现同一角色Q可重复保存Q但应在表中附加说明是由哪个子组导入的。这样在删除子组时就可以有选择性的删除?/p>
但重复子l的情况比较麻烦，q有{考虑。假充有lg01,g11,g12,g21。g01包含g11和g12Qg11和g12分别包含g21。从g01中删除g11Ӟ如何判断g21的去?看来q是应当在维护时判断应不应当删除?/p>
原来GUID只是UUID的一U实现?/p>
原来UUID虽长Q做数据库主键不但不?x)媄响性能Q反而对性能有所帮助?/p>
有时间我实际试一下?/p>
1 Group 3 rows
2 Collapse the group, summary row (row 3#) was collapsed too.
3 However, I can drag and redisplay summary row (row 3#)
4 When I collapse it again, summary row (row 3#) still be collapsed.
Outline of column has same bug. How could I choose OpenOffice?
]]>
]]>
If data is null in DB, and Hibernate maps it as primitive type, An PropertyAccessException will be th rowed when.
Our solution (Discussed with Andy):
1) Ensure all Database fileds have default value.
2) Wrap all primitive value in DTO. (Andy says this is not essential.)
2. Debug Hibernate in my product
In single way one-to-many association, when update, Hibernate always update children's id to null!
Only when Bi-one-to-many association, Hibernate can update children correctly.
Do not use Hibernate List. It need a index/list-index property, which bind to a DB column, but means the real index of Java List. The List often get many null items.
Hibernate set is good. But our DTO already use ArrayList as Collection.
Children Id is missing after updated it.
Hibernate session.save normally call insert, session.update normally call update.
When update Bi-one-to-many association, parent always correct. But always insert children, can not delete and update children. CAUSE: I use a modifiable field as PK, because all children's ID is missing from JSP. In this case, I should manual delete and insert children table.
BTW, Set can not include duplicated items. However. Our DTO have no PK when it created until saved it to DB. This cause add new item fail.
]]>
http://jasperintel.sourceforge.net/
JasperServer is specifically designed to provide a Web application around JasperReports. Continuing the philosophy of JasperReports, JasperServer is aimed at providing a managed, stand alone Web reporting environment, as well as being able to be embedded in existing Java Web applications, and services to a variety of technologies through Web services, including the LAMP stack. By focusing on a clean business API and using an advanced inversion of control architecture, JasperServer can be easily integrated and customized for your needs.
JasperIntelligence is an open source business intelligence platform from JasperSoft, providing common services like security and metadata management, and the capability to easily add additional functionality. The JasperServer is the first service using the JasperIntelligence platform. Following soon will be:
- Report and metadata designers integrated with the platform
- OLAP user interfaces
- Data warehouse management environments
- Data integration services (ETL)
roadmap
JasperServer Planned Features
- Additional report data source types: Hibernate, EJB, POJO, XML
- RTF output of reports through the user interface
- Additional input control types
- Multiple select
- Select from a database lookup - Import and export of repository contents
- Supports moving between repositories ie. test to production - Web services and PHP example application
- Support for PDF, Excel and RTF report renderings - iReport plugin for repository browsing and report deployment
- Includes Web services for repository management - User Search
- Integration of external authentication mechanisms, such as LDAP, JAAS
- Session management without the use of cookies
- Internationalization of the Web user interface
- Management of resource bundles through the user interface
- User Preferences
- Scheduled report generation
- Includes storage of generated content and notifications of new reports to interested users via email - Saving report parameters and rerunning reports using saved parameters
- Additional documentation
- Architecture overview
- Javadoc
]]>